|Home | About | Journals | Submit | Contact Us | Français|
Background Ethnic differences in physical activity in children in the UK have not been accurately assessed. We made objective measurements of physical activity in 9–10-year-old British children of South Asian, black African–Caribbean and white European origin.
Methods Cross-sectional study of urban primary school children (2006–07). Actigraph-GT1M activity monitors were worn by 2071 children during waking hours on at least 1 full day. Ethnic differences in mean daily activity [counts, counts per minute of registered time (CPM) and steps] were adjusted for age, gender, day of week and month. Multilevel modelling allowed for repeated days within individual and clustering within school.
Results In white Europeans, mean daily counts, CPM and mean daily steps were 394785, 498 and 10220, respectively. South Asian and black Caribbean children recorded more registered time per day than white Europeans (34 and 36min, respectively). Compared with white Europeans, South Asians recorded 18789 fewer counts [95% confidence interval (CI) 6390–31 187], 41 fewer CPM 95% CI 26–57) and 905 fewer steps (95% CI 624–1187). Black African–Caribbeans recorded 25359 more counts (95% CI 14273–36 445), and similar CPM, but fewer steps than white Europeans. Girls recorded less activity than boys in all ethnic groups, with 74782 fewer counts (95% CI 66665–82899), 84 fewer CPM (95% CI 74–95) and 1484 fewer steps (95% CI 1301–1668).
Conclusion British South Asian children have lower objectively measured physical activity levels than European whites and black African–Caribbeans.
Compared with European whites, British South Asian adults are at increased risk of type 2 diabetes1,2 and coronary heart disease.3 They show evidence of marked insulin resistance, hypertriglyceridaemia and central adiposity.14 British African–Caribbean adults are at increased risk of type 2 diabetes,4,5 hypertension6 and stroke.7 They show a lesser degree of insulin resistance and raised blood pressure.8–10 There is growing evidence that these differences in disease risk emerge before adult life. British South Asians are at increased risk of type 2 diabetes in childhood and adolescence11 and show evidence of insulin resistance from 10 years of age;12–14 these differences may emerge much earlier15 but this has yet to be established in UK South Asians. Similar, but less marked, differences in insulin resistance are also apparent among black African–Caribbeans.14
Physical inactivity is a risk factor for cardiovascular disease and type 2 diabetes; its metabolic effects are apparent from childhood.16–18 However, the degree to which physical activity patterns differ between British South Asians, black African–Caribbeans and white Europeans remains uncertain. The studies so far conducted have been based on self-reported physical activity levels and suggest that physical activity levels may be lower in South Asian adults;19 limited data suggest a similar pattern in children.20 Few data are available for British African–Caribbeans. However, self-reported physical activity may provide unreliable comparisons between ethnic groups;19 no assessment of physical activity levels in children of different ethnic groups using objective methods of physical activity assessment has been reported.20 We have therefore made objective measurements of physical activity levels using the Actigraph GT1M, a validated uniaxial movement sensor, in a school-based study of the cardiovascular health of 9–10-year-old British children of South Asian, black African–Caribbean and white European origin. We report on overall physical activity levels, time spent at different intensity levels and on the proportion of children aged 9–10 years meeting current physical activity recommendations of at least 1h of at least moderate intensity physical activity per day,21 as well as ethnic and gender differences in these levels of activity.
The Child Heart Health Study in England (CHASE Study) was designed to investigate the health of children of white European, South Asian and black African–Caribbean origin. Ethical approval was obtained from the relevant Multicentre Research Ethics Committee. Information on the ethnic composition of state school populations was obtained from the UK Government Department for Education and Skills. The school sampling frame included all state primary schools in London, Birmingham and Leicester with between 15 and 50% pupils of White European origin. Two separate random samples, each of 100 schools, were taken. The first included schools with a high prevalence (30–80%) of South Asian pupils, stratified by Indian, Pakistani and Bangladeshi origin. The second included schools with a high prevalence (30–80%) of African–Caribbean pupils, stratified by African and Caribbean origin. Head Teachers were approached personally by the Principal Investigator (PHW) and invited to participate. Schools that did not agree to take part (30%) were replaced by a school from the sampling frame with a similar ethnic composition and in the same or a neighbouring borough. Depending on the school size, one or two classes of pupils in Year 5 were invited to participate. Invitation letters (translated where necessary) were sent to parents or guardians of pupils. Schools were visited by the Research Co-ordinator, who explained details of the study to pupils (with parents in attendance if possible). The present report is based on the final phase of the study (January 2006–February 2007, 78 primary schools) during which objective physical activity assessments were carried out.
All assessments were carried out during school terms by a single field team, which visited schools in North-West London, North-East London and South London in rotation, with periodic visits to Birmingham and Leicester. Participating children had physical measurements (anthropometry, blood pressure and spirometry) and provided a fasting blood sample. They were asked to wear an Actigraph GT1M activity monitor (ActiGraph, LLC, Pensacola, FL, USA), which measures uniaxial acceleration 30 times per second, with the counts summed over a user specified time interval (we used 5s) for a defined period (up to 7 complete days where available). Children were shown how to position the monitor (worn at the waist above the left hip using an elasticized belt) and were asked to wear the instrument during waking hours from the time of the survey and then for 7 complete days, returning the instrument to school on the morning of the eighth day (or the next school day if a Saturday or Sunday). Participants were also asked to complete a questionnaire when returning the instrument, to establish whether the monitor was worn every day, whether the child had participated in exercise requiring monitor removal (e.g. swimming) or poorly detected by the monitor (e.g. cycling). Participants were also asked to record illness during the week, and self-perceived level of exercise during the week (usual, more than, less than usual). To encourage return of the Actigraph instrument, a gift voucher was issued on its safe return.
Parents provided information on their self-defined ethnic origin and that of the child and also provided information on occupation (coded using the SOC2000 occupational classification) and educational attainment. Children provided information on parental and grand-parental place of birth. Ethnicity in this report was based on parental self-defined ethnicity for both parents, or (where not available) parentally defined child ethnicity, or (where not available) child-defined geographic origin of parents and grandparents. Children of unmixed ethnicity were classified into three major ethnic groups (i.e. white European, South Asian, black African–Caribbean) with all other children (including those of mixed ethnicity) allocated to an ‘other ethnicity’ group.
ActiGraph data files were downloaded and batch processed using a dedicated software programme MAHUFFE (available from http://www.mrc-epid.cam.ac.uk/). The main measures of overall activity included mean daily activity counts, mean daily steps, and activity counts per minute (CPM) of registered time. Mean daily times (minutes) spent in sedentary (defined as <100CPM), light (100 to <2000CPM), moderate (2000 to <4000CPM), vigorous (>4000CPM) and moderate to vigorous physical activity (MVPA) were also recorded, the latter by combining moderate and vigorous levels of physical activity into one single variable. The cut-off for moderate exercise of 2000CPM is equivalent to walking 4km/h.22–24 Registered time was defined as the total period accepted for analysis; time periods of at least 20 consecutive min of zero counts were deemed to represent periods when the monitor was not worn and were excluded. The incomplete first day and the last morning of recording were omitted. Subjects with a registered time of ≥600min on ≥1day were included in the main analysis; only days with >600min of recorded time were included. The effect of restriction to subjects with 3 days of 600min of registered time was examined. Children who spent ≥60min/day in moderate or vigorous activity (in accordance with Department of Health guidelines) were identified.21
Statistical analyses were carried out using STATA/SE software (Stata/SE 9.2 for Windows, StataCorp LP, College Station, TX, USA). Outcome variables, including mean daily counts, steps, CPM, registered time, and time spent in different levels of activity (except vigorous activity), all appeared to be reasonably normally distributed. For counts, CPM and steps, values outside 3.5 SDs of the median were identified as outliers and removed from the analysis. The SD was calculated by regression of the outcome variable on its associated z-score between the 10th and 90th percentile of the data. Multilevel linear regression models taking account of the natural clustering of children within school and repeated days within individuals were used to provide adjusted means, ethnic and gender differences in physical activity. All analyses adjust for age in quartiles, gender (combined analysis only), day of the week, month and day order of recording. The effect of adjusting time spent in light to vigorous activity for registered time was also examined. Likelihood ratio tests were used to establish whether ethnic differences were modified by sex. Logistic regression models were used to establish the odds of spending ≥60min in at least moderate levels of activity on the first day of recording in different ethnic groups. Using the first day of recording was considered to provide a fair comparison. The odds on subsequent days (amongst those with available data on the second and third day of recording) and the effect of adjustment for registered time were also examined.
Of 3449 children invited, 2393 (69%) participated and 2144 (62%) wore an Actigraph monitor. Of these, 2071 recorded ≥1day with >600min of registered time with no outlying values for counts, CPM or steps; 1841 children recorded at least 3 full days. On average, data were recorded for 5 days from each child. The mean age of participants was 9.9 years (95% reference range 9.2–10.7); 52% were girls. Participants included 562 white Europeans (mainly white British, with small numbers of white Irish and Eastern Europeans), 494 South Asians (36% of Bangladeshis, 35% of Pakistanis and 24% of Indian origin), 607 black African–Caribbeans (60% of African and 35% of Caribbean origin) and 408 children from other ethnic groups (including those of mixed ethnicity). Participation rates did not differ appreciably between the main ethnic groups (63% white Europeans, 66% South Asians, 60% black African–Caribbeans, 61% other; test for heterogeneity, P=0.11). There was no evidence to suggest any ethnic differences in the number of days from which data were recorded (with a mean of 5 days for each ethnic group). For all children, mean daily activity counts, CPM and daily step counts (SD) were 394871 (139324), 482 (167) and 9804 (3508), respectively. Mean registered time per day was 827 (134) min (Table 1).
Mean daily activity counts increased from Monday to Friday and were lower at weekends, especially Sunday; a similar pattern was seen in boys and girls (Figure 1a), and for daily step counts (Figure 1b). However, registered time was higher on Fridays compared with other days of the week (Figure 1c); CPM (which standardizes activity counts for registered time) showed the highest levels on Wednesday to Thursday (Figure 1d). Activity counts, CPM and step counts were higher in summer months than in winter months (Figure 2); patterns were similar in boys and girls. Registered time showed no appreciable seasonal variation. Activity counts, CPM, step counts and registered time all showed a slight decline between the first and third days of recording, with little appreciable change thereafter; patterns were similar in boys and girls (see Figure S1, available as Supplementary data at IJE online). These patterns did not differ appreciably between children of different ethnic groups.
Within the narrow age-group studied, age was not related to any of the physical activity measures (data not shown). Mean activity counts, registered time, CPM and steps in each ethnic group are summarized in Table 2 both separately for boys and girls, and then with genders combined, in analyses that are adjusted for age, circumstances of measurement and gender (combined analyses only); the corresponding differences between ethnic groups (South Asians–white Europeans and black African/ Caribbeans–white Europeans) are presented in Table 3. Children in the ‘other ethnicity’ group had similar activity patterns to white Europeans; because of their heterogeneous composition they are omitted from further ethnic comparisons. In white Europeans (for genders combined), mean daily activity counts, CPM and step counts were 394785 [95% confidence interval (CI) 385501–404068], 498 (95% CI 487–510) and 10220 (95% CI 10001–10438), respectively; mean registered time per day was 799min (Table 2). Compared with white Europeans, South Asians recorded 18789 (95% CI 6390–31187) fewer counts, 41 (95% CI 26–57) fewer CPM and 905 (95% CI 624–1187) fewer steps, despite having 34 (95% CI 21–47) min more registered time per day (Table 3). South Asian children also recorded fewer minutes in moderate and vigorous levels of activity and more time in sedentary activity (Table 3); these differences were similar after adjustment for registered time. In parallel analyses, black African–Caribbeans recorded 25359 more daily counts (95% CI 14273–36445) and 305 fewer steps (305, 95% CI 53–556). They also wore the monitor for, on average, 36min longer (95% CI 24–47); CPM were similar to white Europeans (test for difference P=0.14). Black African–Caribbeans recorded more time in sedentary activity, and more time in moderate and vigorous levels of activity (Table 3). The differences between black African–Caribbeans and white Europeans in time spent in light to moderate levels of activity were weakened after adjustment for registered time, but remained statistically significant for vigorous levels of activity (P<0.0001). Restriction of the analyses to 1841 children with at least 3 complete days of data or more did not alter the overall pattern of differences.
In all ethnic groups combined, girls recorded less activity than boys with 74782 fewer counts (95% CI 66665–82899, P<0.0001), 84 fewer CPM (95% CI 74–95, P<0.0001) and 1484 fewer steps (95% CI 1301–1668, P<0.0001) per day (Table 4). On average, girls wore the monitor for 14min less (95% CI 6–23min, P<0.001) per day compared with boys (Table 4). Girls recorded more time in sedentary activity and less time in light, moderate and vigorous activity. Although differences between ethnic groups appeared stronger in boys compared with girls, there was no strong evidence that ethnic differences varied between genders (all tests for gender–ethnic group interactions; P>0.3).
We examined the time that children spent on moderate to vigorous activity on the first day of recording. Overall, 64% (n=1328/2071) of the study population spent ≥60min/day in moderate to vigorous activity. A lower proportion of South Asians (54%) achieved 60min of moderate to vigorous activity compared with White Europeans (70%) and Black African–Caribbeans (69%); odds ratio for South Asians vs white Europeans 0.60, 95% CI 0.44–0.82, P=0.001. These ethnic differences were not materially affected by adjustment for registered time, or when further days of recording were considered. A lower proportion of girls (53%) spent ≥60min/day of moderate to vigorous activity than boys (76%); odds ratio 0.30, 95% CI 0.25–0.37, P<0.001. There was little evidence to suggest that these gender differences were modified by ethnic group (test for gender–ethnic interaction, P>0.7).
Patterns of physical activity among South Asians were similar among Indians, Pakistanis and Bangladeshis, and those among black African–Caribbean origin were similar among Africans and Caribbeans. The differences between South Asians and Europeans were not affected by adjustment for parental occupational social class (data not shown). The results were not materially affected by exclusion of the small numbers of children who reported cycling (which may not be adequately measured by accelerometry) and swimming (when the monitor was taken off).
This study provides the first evidence using an objective assessment method that British South Asian children have lower levels of overall physical activity (counts, CPM and steps) compared with white Europeans, reflecting longer periods spent in sedentary activity and less time in moderate and vigorous activity. In contrast, black African–Caribbeans have very similar patterns of overall activity to white Europeans, though they appear to spend somewhat more time in vigorous activity. Marked gender differences in physical activity were observed both in the whole study population and within each ethnic group, with girls showing markedly lower levels of overall activity (counts, CPM and steps) than boys, reflecting smaller amounts of time spent in light, moderate and vigorous activity. Consistent ethnic and gender patterns are also seen in the prevalence of children doing moderate to vigorous intensity exercise for at least 60min/day.
White European children in the present study had activity levels (assessed by CPM) that were approximately a quarter lower, in both genders, than those in children of the same age measured with Actigraphs in Denmark, Estonia, Portugal and Sweden.25–27 Activity levels were ~10% lower than in slightly older children (11 years) in the UK-based Avon Longitudinal Study of Children and Parents.28 Differences in registered time are unlikely to account for these activity differences. A number of studies using self-reported physical activity in UK children have suggested that South Asian children may be less active, though these have not quantified differences in overall levels of physical activity.19,20,29–31 Information concerning levels of physical activity in African–Caribbean children compared to other ethnic groups is scarce.31
A particular strength of the present study is the use of the Actigraph monitor to provide objective and validated32 assessment of physical activity, which is particularly important for valid ethnic group comparisons.19 The study provided broadly representative samples of children of South Asian and African–Caribbean ethnicity, and was designed to allow ethnic group comparisons to be carried out on a within-school basis, limiting the influence of confounders operating on a between-school basis (particularly area, social circumstances and season). We have included all available data (even single days for a small proportion of participants) and have adjusted for all factors related to measurement, in particular season, day of week and day order effects in analysis. In contrast, previous studies have required data for ≥3 days of recording and did not adjust for the season, day of week or day order.25,26,28 Our approach allows more subjects to be included in the analysis (2071 in the analysis based on a single day, compared with 1841 children in the analysis based on ≥3 days), and maximized participation rates, limiting the potential for selection bias. However, the response rate for the study was modest (69%, with 62% of children providing adequate Actigraph data). It is possible that a degree of selection bias, likely to reflect lower participation rates among less active children and leading to overestimation of true levels of physical activity, is present. However, the demographic, ethnic and anthropometric characteristics of participating children who did and did not provide Actigraph data were similar (data not shown). This is therefore unlikely to invalidate the ethnic group comparisons described in the report.
Our finding that registered time differs between ethnic groups, being greater in both South Asians and black African–Caribbeans compared with white Europeans, raises important issues about the standardization of comparisons of accelerometer data. Previous studies have reported CPM of registered time on the assumption that belts were not worn during unregistered time and that the activity during that time occurs at the same rate as during registered time. These are major assumptions, which we have previously found to be inappropriate in older adults wearing accelerometers.33 It is therefore reassuring that our finding of lower activity levels amongst South Asians is present for activity counts, CPM and steps (being 5, 8 and 9% lower, respectively). Black African–Caribbean children recorded 6% more counts and a similar CPM, but 3% fewer steps than white Europeans. These discrepancies suggest that deriving steps from these instruments, while appropriate in older adults,33 may not fully describe the diversity of physical activity undertaken by children. Whereas activities in the elderly are predominantly walking based,33 data from this study suggest that in addition to walking fast, running/jogging and chasing games, football in boys and dancing in girls are the main activities reported in these children. Further work to investigate individual and cultural differences in registered time, and how these objective measures relate to the types of physical activity undertaken in childhood is needed.
The finding that South Asian children had consistently lower levels of all activity outcomes including total counts, CPM and steps compared with white Europeans (despite recording a longer average period of registered time), supports the contention that children from this ethnic group have lower levels of physical activity than white Europeans. These lower levels particularly appear to reflect the lesser periods of time spent by South Asians in MVPA and the longer periods spent in sedentary levels of activity. Although black African–Caribbeans recorded more activity counts and recorded longer periods in higher levels of activity than white Europeans, they recorded longer average registered times. There was no evidence of differences in overall activity levels once the difference in registered time was accounted for. However, the longer duration of higher intensity activity amongst black African–Caribbeans remained statistically significant after adjustment for registered time.
For children and young people, a total of ≥60min of at least moderate activity each day is recommended (and at least twice a week this should include activities to improve bone health, muscle strength and flexibility).21 In this study, moderate levels of activity were defined as 2000 CPM, equivalent to walking 4km/h.22–24 While this cut-off is comparable with those of other studies,22–24 it has recently been suggested that higher thresholds of >3000CPM could be used.34,35 The effect of using this higher cut-off is unlikely to materially affect South Asian–white European or gender comparisons (these differences are apparent at both 2000 and 4000 CPM), though the absolute percentage of those achieving these higher cut-off will be lower, reducing the statistical power of ethnic comparisons. Using the present cut-off of 2000 CPM, only 64% of children in this study recorded this intensity of activity, with fewer (54%) achieving this amongst South Asians. This compares well with self-reported levels of physical activity, where in the 2004 Health Survey for England, 65% of 2–15-year olds in the general population reported exercising at recommended levels, with a lower proportion achieving this level of activity (just under half) in South Asians.30 Similar to our findings based on objective measurements, higher self-reported levels of physical activity in white and black Caribbean children compared with South Asians were observed in the 1999 Health Survey for England.31 However, gender differences in self-reported levels of activity (with 69% of boys and 61% of girls reporting recommended levels) were less marked than those measured objectively (76 vs 53%, respectively).30 Due to potential gender/ethnic biases in reporting of physical activity, objective measures offer a more reliable form of assessment.
Previous studies have shown that low levels of physical activity in childhood, measured by accelerometry, are associated with greater adiposity, insulin resistance and adverse cardiovascular risk factor levels.17,25–27,36,37 British South Asians have increased risks of type 2 diabetes, insulin resistance and impaired glucose control in childhood and adolescence.11–14 The lower levels of overall physical activity among South Asians (5–10% lower on the various indices) and lesser amounts of time in moderate to vigorous activity (~7% lower among South Asians) could be making an important contribution to these metabolic differences. The extent of this contribution, and the scope for reversal by increasing physical activity levels among South Asians, will be important issues for further investigation. In contrast, although emerging evidence of insulin resistance and hyperglycaemia is also apparent in black African–Caribbean children when compared with white Europeans,14 differences in physical activity are not likely to be a major contributing factor.
Overall physical activity levels among British South Asian children are lower than those among white Europeans and black African–Caribbeans. Understanding the reasons for the ethnic difference and its implications for the increased risks of cardiovascular disease and type 2 diabetes among British South Asians is an important priority.
Wellcome Trust (068362/Z/02/Z); the British Heart Foundation (PG/06/003); National Prevention Research Initiative (NPRI). The Funding Partners for this NPRI award were: British Heart Foundation; Cancer Research UK; Department of Health; Diabetes UK; Economic and Social Research Council; Medical Research Council; Research and Development Office for the Northern Ireland Health and Social Services; Chief Scientist Office, Scottish Executive Health Department; and Welsh Assembly Government.
Supplementary data are available at IJE online.
We are grateful to the members of the CHASE study team (Julie Belbin, Angela Brock, Claire Brannagan, Sarah Holloway, Cathy McKay, Mary McNamara, Miranda Price, Rahat Rafiq, Chloe Runeckles, Lydia Shepherd, Andrea Wathern) and to all participating schools, pupils and parents.
Conflict of interest: None declared.